Measurement of the serum vitamin B₁₂ level has been widely used as the standard screening test for cobalamin deficiency in various demographic and disease groups. Despite its overall utility, the test has poor positive and negative predictive values because of formidable problems of sensitivity and specificity [3]. A low serum vitamin B₁₂ level frequently does not indicate cobalamin deficiency, and a "normal" serum vitamin B₁₂ level does not reliably indicate normalcy. The distribution of circulating cobalamin among its several plasma-binding proteins is one reason for this problem. Most plasma cobalamin is bound to transcobalamin I (TCI). The function of this binding protein is unknown, but it appears to play little, if any, role in the cellular delivery of the vitamin [4]. The minor fraction (10% to 20%) of plasma cobalamin that is bound to a second protein, transcobalamin II (TCII), is the functional and clinically important component. Patients with congenital deficiency of TCII have features of severe cobalamin deficiency but usually have normal serum vitamin B₁₂ levels [5]. Conversely, when TCII-bound cobalamin (also called holo-TCII) levels are low but the amount of cobalamin bound to the TCI fraction is increased, deficiency may not be detected if only the total serum vitamin B₁₂ level is measured. Plasma levels of TCI, which originates in granulocytes, increase in myeloproliferative disorders; thus, the rare patients with coexistent chronic myelogenous leukemia and pernicious anemia have normal total serum vitamin B₁₂ levels [6, 7]. Some clinical evidence supports the concept that measurement of holo-TCII levels may provide a better index of cobalamin status [8, 9], but reliable commercial screening assays for holo-TCII are not yet available.

An alternative approach to directly measuring serum vitamin B₁₂ levels has recently become available for identifying cobalamin deficiency. Cobalamin is required as a cofactor for two enzyme-mediated biochemical reactions that become abrogated during deficiency, with consequent pile-up of the two metabolites, methylmalonic acid and homocysteine. Clinical diagnostic assays for measuring these compounds in the serum and urine are now available. Metabolite assays seem to provide greater sensitivity and specificity than does direct measurement of blood vitamin levels [10, 11]. However, some concerns about the use of these assays and the interpretation of their results are emerging [12]. Metabolite levels also may be elevated in many other inherited and acquired disorders. This topic has recently been reviewed in detail elsewhere [3]. Impaired renal function results in a nonspecific elevation of both plasma methylmalonate and homocysteine levels and is perhaps the most common problem. Measurement of urine methylmalonate levels, corrected for creatinine, obviates this difficulty [13]. Other confounding variables seem to affect homocysteine levels alone. Elevations are seen with folate and pyridoxine deficiencies, hypothyroidism, and several inherited defects affecting the two major pathways of homocysteine metabolism. Sumner and associates [1] did not study possible causes of elevated metabolite levels other than folate levels. Achlorhydria and the intestinal blind loops resulting from gastrectomy may provide sanctuaries for bacterial overgrowth. In addition to contributing to cobalamin deficiency through microbial gluttony for vitamin B₁₂ [14], the bacteria may also provide another source of methylmalonate. Bowel sterilization with antibiotics can result in decreased serum methylmalonate levels [15].

With no clear, independent, objective (and correctable) clinical evidence of cobalamin deficiency, the demarcation between the lower end of normalcy and true early biochemical cobalamin deficiency remains blurred. This is true of assays for serum vitamin B₁₂ and assays for metabolites. The problems of diagnosing subtle and atypical forms of cobalamin deficiency have recently been addressed elsewhere [7, 16]. In the study by Sumner and colleagues, the standard measurable hematologic complications of cobalamin deficiency either were absent or might have been obscured by concomitant iron deficiency, which was not directly evaluated. A frequent hematologic complication of gastrectomy [2, 14], iron deficiency, not only causes anemia but also masks macrocytosis [17]. In Sumner and colleagues' study, treatment with vitamin B₁₂ decreased serum metabolite levels; this finding is similar to that of earlier studies of elderly persons with increased methylmalonate but normal serum vitamin B₁₂ levels [18, 19]. Whether this confirms the existence of preclinical cobalamin deficiency is moot in patients with no objective evidence of deficiency or measurably improved clinical status. These patients may have early, biochemical cobalamin deficiency that would ultimately progress to evident disease.

The question of the prevalence of cobalamin deficiency in the general population is also vexing. Discrepancies among reports on demographically similar groups raises questions about the standardization of metabolite assays and the definition of normal reference ranges. In Sumner and colleagues' study, for example, the low 2% prevalence of cobalamin deficiency found in the elderly controls contrasts markedly with the 14.5% prevalence of deficiency reported in elderly persons on the basis of metabolite assays [18]. What makes this difference even more remarkable is that the assays for both studies were done in the same laboratory, apparently using the same method. Furthermore, in the earlier study, a more liberal definition of normalcy was used to assess cobalamin status. The investigators of that study defined the upper limit cut-point of a broad normal range as 3 standard deviations above the mean; Sumner and colleagues used only 2 standard deviations as the cut-point. More rather than fewer patients with apparent "biochemical" cobalamin deficiency would have been expected among the controls in Sumner and colleagues' study than in the earlier study. It seems unlikely that the suggestions of Sumner and colleagues—that the slight age differences between the groups—and the possibility that some participants in the earlier study might have had gastric surgery could completely explain the striking differences in the apparent prevalence of cobalamin deficiency.

What, then, is the current place of vitamin B₁₂ and metabolite assays in diagnosing cobalamin deficiency and managing persons at risk for developing this deficiency, especially the elderly and patients who have had gastrectomy? Indiscriminate use of metabolite assays as a primary screening test currently seems unjustified because of unresolved questions about the tests and their cost. Sumner and coworkers [1] conclude that in patients who have had gastric surgery, regular and continued monitoring of the serum vitamin B₁₂ level is an adequate screening test for cobalamin deficiency. Patients with only a partial stomach are also susceptible to the inexorable process of age-related diminution of gastric function that leads to a higher prevalence of cobalamin deficiency among normal elderly persons. The rate of erosion of body cobalamin stores in patients who have had gastrectomy may accelerate, however, and vigilance for the onset of deficiency must therefore be maintained indefinitely. In groups at high risk for developing cobalamin deficiency (including persons who have had gastrectomy and the elderly), it is reasonable to resort to metabolite assays to resolve discrepancies between the serum vitamin B₁₂ level and the clinical and hematologic profile. A normal serum vitamin B₁₂ level in a patient suspected of having cobalamin deficiency or a low serum vitamin B₁₂ level without convincing evidence of deficiency certainly justifies measurement of metabolite levels. For cost considerations, simply treating any high-risk patient who has a low serum vitamin B₁₂ level is also reasonable. With the minor and rare side effects of vitamin B₁₂ treatment, safety is hardly an issue; however, long-term compliance is. The availability of metabolite assays for studying patients with cobalamin disorders or deficiency is potentially useful, but until unresolved issues have been settled, the assays should be used judiciously. In a cost-conscious era, buyers of expensive specialized laboratory tests must be selective—and wary.